From infancy to young adulthood: Exploring the divergent genomic mechanisms that drive AML.

Abstract

6544 Background: Pediatric & young adult Acute Myeloid Leukemias (pAML & YA AML) are an uncommon, heterogenous, & clinically challenging group of malignancies, as age ranges widely & proper therapy selection (i.e. pediatric vs. adult regimens, etc.) can be unclear. The genomics of these disorders can help understand distinct disease biology, better risk stratification, & improve outcomes. Methods: Bone marrow, peripheral blood, or FFPE samples from 870 suspected AML patients were sequenced using a 302 gene panel to detect SNVs/indels. 466 underwent RNA fusion detection of 184 genes & DNA analysis for CNV detection of 24 genes. Only pathogenic/likely pathogenic variants were included in the analysis. AML patients were split into 4 groups based on age: Infant (≤3, n=30), Childhood (4-14, n=51), Adolescent (15-19, n=51), YA (20-35, n=738). 4750 DNA/RNA sequenced adult AML patients (>35) were also included in the analysis. Statistics were performed using Fisher’s exact test. Results: pAML patients (infant, childhood, & adolescent) had a high prevalence of FLT3 & NRAS variants (17.4% & 15.2%) & fusions (26.2%) while YA AML had a high prevalence of fusions (29.2%), FLT3 (20%), NRAS (12.5%), & WT1 (12.6%) variants. Adult AML had a high prevalence of TET2 (18.4%), DNTM3A (18.3%), ASXL1 (16.8%), SRSF2 (14.8%), & TP53 (14.2%) variants, highlighting an increased prevalence of mutations associated with clonal hematopoiesis or prior chronic myeloid neoplasms. Infant AML had a higher number of GATA1 variants (33.3% vs. 0% - 2%, p<0.00001) vs. childhood, adolescent, & YA AML; trisomy 21 was present in 90% of patients by CNV detection. Fusions were only found in 5% of infant AML (p=0.02) vs. 29.2% - 34.5% in other groups. FLT3 variants were lower in infant (10%) and childhood AML (13.7%) vs. adolescent (25.5%) & YA (20%) while NRAS variants were higher in adolescent AML (21.6% vs. 6.7% - 13.7%). Childhood AML had a higher prevalence of RUNX1 fusions (16.1% vs. 0 – 5.4%, p=0.03) & Adolescent AML had a higher prevalence of PML::RARA fusions (13.8% vs. 0% – 6.5%) vs. other groups. YA AML had a higher prevalence of WT1 variants (12.6% vs. 0% - 7.8%, p=0.02). CNV loss in IKZF1 (7p12) were more prevalent in adolescent AML (10.3% vs. 0-0.5%, p=0.002); EZH2 CNV loss (7q36) were more prevalent in YA & childhood AML (4.7% & 3.4% vs. 0%). Upon aggregating genes by their molecular function, infant AML had a lower prevalence of variants in epigenetic genes (6.7% vs. 15.57-24.8%, p=0.02), RAS (10% vs. 23.2%-25.5%), & signaling genes (16.7% vs. 27.5% – 41.2%). Variants in DNA repair genes were more frequent in adolescent AML (13.7% vs. 3.9% - 6.7%, p=0.01). Conclusions: Infant, childhood, adolescent, & YA AML harbor unique genetic profiles that distinguish themselves from each other and reflect divergent and evolutionarily favored mechanisms behind leukemogenesis. Understanding these genetic profiles can help predict prognosis & help tailor more effective treatments.